We are potentially migrating from the Vapi SDK because of this one use-case that fails:
Agent calls IVR-gated number
Agent navigates the IVR
Agent is transferred to selected number
Agent should be able to continue the conversation post-call.
In practice, we’ve seen the agent fail to pick up audio sporadically on these transfers. It’s hit and miss. Has anyone built a similar workflow in Pipecat/ is it production-reliable?